6th International Conference on Earthquake Geotechnical Engineering 1-4 November 2015 Christchurch, New Zealand

Liquefaction Resistance of Christchurch CBD soils

M. L. Taylor1, M. Cubrinovski2, and B. A. Bradley3

ABSTRACT The liquefaction-susceptible soils that underlie the Christchurch CBD have been evaluated within a

comprehensive field (CPT, borehole, shear-wave velocity) and laboratory-testing programme. Sampling sites exhibited extensive ground damage following the earthquakes, requiring demolition of multi-storey commercial buildings. Liquefaction resistance was evaluated by direct measurement using cyclic triaxial testing on undisturbed specimens obtained from 2 to 14 m depth of CBD deposits, using novel Gel Push (GP) sampling. Tested samples cover a broad range of soils from clean sands (SP) to silty sands (SM), sandy silts (ML), with fines contents (typically non-plastic) ranging from 1 – 98 %. Cyclic strength testing allows comparison to the empirical correlations based on case-history data, which now include case histories from the recent events in Christchurch, New Zealand. While generally good agreement is obtained for clean sands, comparison of sands with high non-plastic fines indicate more research is warranted to clarify the cyclic strength of these deposits, given their prevalence in Christchurch.

Introduction

Following the 2010-2011 Canterbury Earthquake Sequence (CES) a research investigation was conducted to better characterise the deposits most at risk of soil liquefaction. Two sites in the Christchurch Central Business District (CBD) that exhibited extensive liquefaction adversely impacting building foundation performance were targeted for detailed investigations: CPT testing; borehole drilling; shear wave velocity (Vs) testing; and Gel Push (GP) sampling (Bray et al. 2014, Taylor et al. 2012). GP is a new method for obtaining undisturbed specimens of sandy soils from below the water table, and trialled in Christchurch following promising work by Kiso Jiban Consultants (Japan), and Huang et al. 2008 (Taiwan). The GP samples were subjected to cyclic (CTX) and monotonic triaxial loading, and compared to a suite of tests on reconstituted specimens of representative GP samples. This comparison identifies the influence of natural features (structure, fabric) on the soil response. The lab results, coupled with field profiling data, and complimented with soil index properties (gradation, specific gravity (Gs), Atterberg limits, max & min void ratio (emax, emin), and grain shape) provide a comprehensive soil characterisation. This paper summarizes the project, some of the key findings, and proposed future research directions.

Geology of Christchurch CBD soils

The Christchurch CBD was founded by European settlers (1850) on the nearest high-dry land above the coastal swamps suitable for establishing the township (Wilson 1989). It is described as an “island” of gravel within the swamp, the easternmost extent of a legacy flood channel of the Waimakariri River. Either side of these gravel channels are overbank deposits of sand and silt, with clays/ organics 1Geotechnical Engineer (ARUP) & Research Engineer (Univ. Canterbury), Christchurch, NZ, merrick.taylor@canterbury.ac.nz 2Professor, Dept. Civil Eng., University of Canterbury, Christchurch, New Zealand, misko.cubrinovski@canterbury.ac.nz 3Assoc. Prof., Dept. Civil Eng., University of Canterbury, Christchurch, New Zealand, brendon.bradley@canterbury.ac.nz

found further from the recent channels. Between flood events, the spring-fed Avon River meanders west-to-east through the city, developing large meander loops on its way to the estuary and Pacific Ocean.

Post-quake reconnaissance and ground investigations While evidence of minor sand boiling and surface water from liquefaction were commonly encountered across Christchurch including the CBD, observations of extensive and severe soil liquefaction were isolated to particular regions of the CBD (Cubrinovski et al. 2011), notably adjacent to recent gravel channels that incur into the CBD, where flood overbank deposits of sand and silty sands have accumulated without capping by a low permeability crust. This includes Kilmore Street east of Colombo, Salisbury and Peterborough Streets east of Manchester, and within the Avon Loop, south of the Avon River between Barbadoes and Fitzgerald Streets (refer Figure 1). A smaller area of extensive liquefaction was also observed to the south of the recent gravel

channel following the present Avon River course, on the corner of Madras and Armagh Streets. Extensive, severe and repeated liquefaction was observed in these areas affecting multi-storey

buildings (piled and shallow founded), requiring demolition post-quake. Both the Kilmore and the Madras-Armagh areas were targeted for intensive CPT testing, boreholes and trial GP sampling (Bray

et al. 2014). The typical soil profile from Kilmore Street, adjacent to Transport House is shown in Figure 2, showing the thick profile (2 – 8 m) of silty-sands overlying clean sands.

Gel-push sampling and specimen quality

Cyclic testing of undisturbed samples provides a measure of in situ response, as soil ageing, fabric and layering can be significant (Ishihara, 1993). GP sampling was selected to obtain undisturbed samples

from below the water table without resorting to expensive ground freezing. Figure 3a) depicts the GP-S (piston-type) sampling procedure, while Figures 4b) and 4c) show the sampling trial in the CBD.

Figure 1: Christchurch CBD map with liquefaction observations and location of sampling sites (yellow stars). Photos: A) Transport House, 151 Kilmore Street, (B) Markhams Building, 144 Kilmore Street;

Moorhouse Ave.

Bealey Ave.

Dea

ns A

ve.

Fitz

gera

ld A

ve.

Rol

lest

on A

ve.

Hagley Park

Christchurch Hospital

Avon River

200 m

1000 ft

N

Col

ombo

Str

eet

Man

ches

ter

Str

eet

Kilmore Street

Mad

ras

Str

eet

Armagh Street

Bar

bado

es S

tree

t

Mon

trea

l Str

eet

Dur

ham

Str

eet

Salisbury Street

Peterborough Street

Oxfor

d Terr

ace

Camb

ridge

Terra

ce

A. B.

C.

Christchurch CBD and sampling sites

Site K1

Severe Liquefaction Moderate Liquefaction Minor/ No Liquefaction

Site MA1

(C) Amicus House, 240 Armagh Street (near) and Trade Union Building, 199 Armagh Street (far).

Figure 2: Soil profile at site K1, on Kilmore Street between Colombo and Manchester Streets.

For details of GP-S trials and initial sample quality appraisal refer Taylor et al. (2012a), a detailed evaluation is presented by Taylor (2014). Generally good quality silty-sand samples were obtained, with well-preserved soil fabric and structure (refer Figure 4E), with a close match between field and lab-measurements of Vs. The Vs test results also indicate that ageing effects are likely to have been erased/ reset during the large shearing induced during the recent quakes, or that they were not significant prior to the quakes (McGann et al. 2014, pp. 20-21). Further evaluation of the GP sampler is ongoing with recent field trials at further sites around Christchurch.

Figure 3: (a) Diagrammatic representation of GP-S sampling downhole (i) insertion at base of drilled hole (ii) advance into virgin soil (iii) closure of core catcher (iv) retrieval of sample. (b) drilling on site

in Christchurch CBD (c) retrieval of GP sampler.

Figure 4A shows GP sample harvesting from sample liners, with SM and SP samples shown (Figure 4B and C resp.). The samples were trimmed from 70mm dia. x 120 high to 50 mm dia. x 100 high and mounted in the triaxial apparatus for testing (Figure 4D, E and F). Note the highly stratified structure of the trimmed SM sample is well preserved indicating high quality (Figure 4E). Fluvial sands of the

MADE GROUND: Sandy gravel

21m

0m

2m

8m

SILTY SAND: Grey, fine silty sand/ silt with fine sand [SM/ ML][SPRINGSTON FM]

MEDIUM SAND: Brown, medium sand. [SP][CHRISTCHURCH FM]

20m SILT-CLAY: Grey silt-clay with fine sand, some peat. [ML/ MH] [CHRISTCHURCH FM]

Dep

th [m

]

0

2

4

6

8

10

12

14

16

18

20

qc1N

0 100 200

SBT Index Ic

1 2 3

% Fines

0 25 50 75 100

Springston Formation (Sp. Fm) exhibited a wide range of fines contents (FC, % passing 75µm sieve) where fines were largely non-plastic silt (

Figure 5A).

Figure 4: A: Harvesting undisturbed samples; B) silty sand (SM) sample; C) clean sand (SP) sample;

D) trimming operation; E) trimmed silty sand sample; F) set-up in triaxial apparatus.

Figure 5 presents the particle size distribution curves (PSD) for samples obtained from K1 site (Kilmore St.), which are fairly uniformly graded. The marine beach sands (

Figure 5B, Christchurch Formation, Ch. Fm.) were clean of fines, having been deposited under high-energy wave loading, and existed in situ at higher densities (DR > 60%). Each GP sample was measured for Vs (using bender elements), emax, emin, and Gs. Atterberg Limit tests targeted soils with some noted plasticity at the Madras-Armagh site.

Figure 5: Particle size distribution curves of GP samples tested from Kilmore St. at depths (Z) noted.

A. B. C.

D. E. F.

(A) (B)

Cyclic Response of GP Samples The results of CTX testing are interpreted by constructing a cyclic strength curve (cyclic stress ratio CSR vs. number of cycles, Nc, to induce ‘liquefaction’, defined by strain-based proxy of 5 % double amplitude axial strain, DA εa). Cyclic strength curves are shown in Figure 6, constructed from 3 - 4 No. CTX tests on GP samples of silty-sand from the same sample tube/ depth. Note the variation in sample FC and qc1N, which is undesirable for understanding the influence of these factors on the cyclic strength requiring further interpretation to normalise the data.

Figure 6: Cyclic strength curves constructed per sample tube. Intercept at 15 cycles (CRR15) indicated.

Interpreted Response of Christchurch Silty Sands

Figure 7A presents all K1 test results compared following correction for the field condition (Cr factor, refer inset for relationship) and for confining stress to 1 atm. using Kσ factor after Idriss and Boulanger (2008). The blue shaded region is for Sp. Fm. silty sands. Despite variations in FC and density, the normalized cyclic strengths of the Sp. Fm. samples were quite uniform, while the clean sands (Ch. Fm.) exhibited higher cyclic resistance. Natural variations in soil gradation; state; layering; and fabric with depth add complexity to

characterising the effects of FC and density on the cyclic strength of GP samples. The adopted approach involved using FC to group GP samples within a geological unit by ‘soil type’, similar to USCS class (SP, SM, ML). Once corrections applied, and results grouped in this manner, alternative cyclic strength curves were constructed (Figure 7B). Selected ‘representative’ samples from each FC range were subjected to multiple tests on moist-tamped (MT) reconstituted specimens. These were also corrected to field CRR15 values. Both sets of results have been compared to the empirical liquefaction-triggering curve (Taylor et al. 2015), with results summarised herein. The test site from Madras-Armagh GP sampling site (MA) provided test results that included soils that had low-plasticity (PI 5-13), and these samples exhibited notably higher cyclic resistance than non-plastic soils (cyclic strength curves not presented in this paper).

Figure 7: A) Comparison all cyclic strength test data for Kilmore Street site, corrected for the field condition and normalised for confining stress. B) Data points when sorted by soil type and density

Comparison to the empirical liquefaction triggering curve

The CPT-based empirical liquefaction-triggering curve is plotted in CRR15, qc1N space. To plot cyclic test data on the plot requires correlation between qc1N and specimen DR, known to be a function of the soil gradation (Cubrinovski & Ishihara 1999). The best quality GP samples (known FC, DR, and adjacent qc1N data) were thus used to derive a soil-specific correlation as a function of FC (Taylor et al. 2015). This allowed for estimates of the equivalent qc1N corresponding to the CRR15 values derived from the cyclic strength curves for similar soil type. By plotting the data on the triggering curves corresponding to the FC of the specimens, an independent check of the empirical curve may be considered for the soil. There is acknowledged a significant degree of uncertainty with the correlations and with the adopted Cr factor, however it is a useful exercise to indicate possible trends or areas requiring further consideration. Plots are presented in Taylor et al. (2015) for each FC range considered (FC 35% (probability of liquefaction, PL 15%, 50% and 85% shown which present the expected range after Boulanger & Idriss, 2012) with data points (triangles) from cyclic strength testing of soils with FC 30 – 50 % range, error bars indicate expected uncertainty in the correlations adopted. The

(A) (B)

results indicate that for high FC silty sands, the position of the test-derived data points fall well below the expected range (i.e. approx. coinciding with the PL of 50% curve, e.g. as shown by Boulanger & Idriss (2014, p.97) for individual clean sands). The exception being one MA data-point (inverted blue triangle) shown that exhibited PI of 8, and fell in the expected range. Of note, a study by Carraro et al. (2003) on Ottawa sand with non-plastic fines (silica flour), showed higher FC resulting in a marked shift of the triggering curve towards the right, which Boulanger and Idriss (2014) noted as being due to drained soil response in the study, whereas undrained behaviour would result in lower qc1N in the field (i.e. contractive response with excess pore water pressure development). While that trend is not what is shown here –with sand with high FC exhibiting higher CRR15 for a given qc1N than clean sands, and the in situ state parameter estimates from CPT in this soil were negative indicating a dilative response, there remains a significant shift to the right from what is expected from the empirical triggering method. Possible reasons include larger uncertainty in Cr than expected, sampling disturbance affecting the adopted correlation to qc1N, as well as limitations of the case-history dataset with regards to characterising the influence of fines on the CPT-based liquefaction triggering curve.

Figure 8: Comparison of collated cyclic test data for Christchurch sands with FC 30-50%, and IB08 liquefaction

triggering curve. PSM and FBM test data from Rees (2010). MT = moist tamped. GP = gel push

Summary and Conclusions An investigation of the cyclic resistance of sands from Christchurch CBD has been conducted using state-of-the-art methods, including field profiling (CPT, borehole, Vs), and high quality undisturbed

clean sand curve (PL 15%)

FC = 35% FC = 15%

PL 15%

PL 50%

PL 85%

PI: 8

PI:

sampling (GP piston sampler). The GP samples are from areas in the CBD significantly affected by liquefaction during the Canterbury Earthquakes 2010-2011. Cyclic triaxial testing and complimentary index testing was conducted and results interpreted for the field condition. The data allows for comparison to the simplified liquefaction-triggering curve, as used extensively in engineering practice. A comparison is presented for soils with high non-plastic fines content. The results from both GP and reconstituted specimens (MT) exhibit lower cyclic resistance than the empirical triggering curve, an exception being a specimen with some noted plasticity. The results suggest that the influence of fines on the liquefaction triggering method is worthy of further investigation –particularly due to the paucity of case history data used to define the triggering curve for soils with high FC and low/ no plasticity. Factors affecting the sample response (sample disturbance, correction to the field condition) may also affect the comparison presented. Research in these matters is ongoing at the University of Canterbury.

Acknowledgements

The authors acknowledge the support of the Natural Hazards Research Platform, EQC and the University of Canterbury. We would also like to thank Mr. Ian Haycock (McMillan Drilling Ltd), Kiso Jiban Consultants, and Mr. Kazuki Horikoshi (Tokyo Institute of Technology).

References Bray JD, Cubrinovski M, Zupan J, Taylor ML Liquefaction Effects on Buildings in the Central Business District of Christchurch. Earthquake Spectra, 2014, 30(1): 85-109

Brown LJ, Beetham RD, Paterson BR, Weeber JH. Geology of Christchurch, New Zealand. Environmental & Engineering Geoscience, 1995, 1(4): 427-488.

Boulanger RW, Idriss IM. CPT and SPT Based Liquefaction Triggering Procedures. Davis, CA., Dept. Civil & Environ. Eng., UC Davis, 2014

Boulanger RW, Idriss IM. Probabilistic Standard Penetration Test-based Liquefaction Triggering Procedure. J. Geotech. Geoenviron. Eng. 2012, 138(10):1185-1195

Carraro, J.A.H., Bandini, P., Salgado, R. Liquefaction Resistance of Clean and Nonplastic Silty Sands Based on Cone Penetration Resistance, J. Geotech. Geoenviron. Eng., 2003, 129(11): 965-976

Cubrinovski M, Ishihara K. Empirical correlation between SPT N-value and relative density for sandy soils. Soils and Foundations, 1999, 39(5): 67-71.

Cubrinovski M, Bray JD, Taylor ML, Giorgini S, Bradley BA, Wotherspoon L, Zupan J. Soil Liquefaction Effects in the Central Business District during the February 2011 Christchurch Earthquake. Seis. Res. Let., 2011, 82(6): 893-904.

Huang AB, Tai YY, Lee WF, Ishihara K. Sampling and field characterization of the silty sand in Central and Southern Taiwan. Proceedings 3rd International Conference on Site Characterization (ISC-3), Taipei, 2008, Taylor & Francis: 1457-1463.

Idriss IM, Boulanger RW. Soil Liquefaction during Earthquakes. EERI: Oakland, CA, 2008.

Ishihara K. Liquefaction and Flow Failure during Earthquakes. Géotechnique, 1993, 43(3): 351-415.

McGann, CR, Bradley BA, Cubrinovski M, Taylor ML, Wotherspoon LM. Development and evaluation of CPT-Vs correlation for Canterbury New Zealand soils of the shallow Christchurch and Springston Formations. Research Report 2014-01, Civil and Natural Resources Engineering, University of Canterbury: Christchurch, NZ, 2014

Rees S, Effects of fines on the undrained behaviour of Christchurch sandy soils. PhD Thesis, University of Canterbury, 2010. 282 p.

Taylor ML, Cubrinovski M, Haycock I. Application of new 'Gel push' sampling procedure to obtain high quality laboratory test data for advanced geotechnical analyses. 2012 NZ Soc. Earthquake Eng. Conf. Christchurch, New Zealand.

Taylor ML, Cubrinovski M, Bradley BA. Characterisation of Ground Conditions in the Christchurch Central Business District. Aust. Geomech. J., 2012, 47(4): 43-57.

Taylor ML, The Geotechnical Characterisation of Christchurch Sands for Advanced Soil Modelling. PhD Thesis, University of Canterbury, 2014

Taylor, M.L., Cubrinovski, M., Bradley B.A. Earthquake-induced Liquefaction triggering of Christchurch sandy soils. 2015 NZ Soc. Earthquake Eng. Conf., Rotorua, New Zealand, 2015

Vaid YP, Sivathayalan S. Static and cyclic liquefaction potential of Fraser Delta sand in simple shear and triaxial tests. Canadian Geotechnical Journal, 1996, 33(2): 281-289.

White PA. Avon River springs catchment, Christchurch City, New Zealand. Aust. J. Earth Sci., 2008, 56(1): 61-70.

Wilson, J. Christchurch Swamp to City. Te Waihora Press for the Christchurch Drainage Board: Lincoln (NZ), 1989.

Main MenuConference ProgrammeAuthor Index